Japan's Pension System -Evaluating the 2004 Reform and Establishing Clear Principles for Further Reforms-

RIETI Faculty Fellows HIGUCHI Yoshio and KUROSAWA Masako made the following report titled "What Further Reforms of the Pension System Can Promote the Labor Force Participation of the Elderly in Japan?"

The two objectives of our research were as follows.

(1) To analyze the impact of past pension system reform on the labor supply behavior of senior citizens and to study a pension system that promotes their participation in the labor force.

(2) To study a desirable employment system that expands employment opportunities in order to absorb an increased supply of senior citizens in the labor market brought about by pension system reform.

Using individuals' survey sheets from the Survey of Working for Old-Age People for the 1992 - 2000 period, we devised a dynamic program to model the impact on the labor market supply of older men caused by the 1995 revision of the Old-Age Pension for Active Workers system and the 2001 revision of the Old-Age Employees' Pension system

Using the results of this model we conducted simulations for changes in the payment of pension benefit and in the pensionable age. As a result, we found that when benefit reduction under the Old-Age Pension for Active Workers is zero, the proportion of unemployed people and part-time employees among senior citizens declines, while the proportion of fulltime employees increases. We also found that if the pensionable age is raised to 65, the proportion of fulltime employees increases. Therefore, it is clear that senior citizens choosing to participate in the labor force have an impact on the economy.

Using individuals' survey sheets from the Survey of Working for Old-Age People in 2000 and 2004, we analyzed the impact of the 2001 revision of the Old-Age Employees' Pension system and the 2002 revision of the Old-Age Pension for Active Workers on the labor supply behavior of senior men based on an inductive labor supply model. We observed that raising the age for receiving the basic portion of the Special Payment of Old-Age Employees' Pension, a result of the 2001 revision of the Old-Age Employees' Pension system, had the effect of increasing the labor supply of senior citizens around 60 years old.

Based on the Labor Force Survey we examined recent changes in the employment of men in their early sixties. The labor population and the employment rate of senior men (aged 60 - 64) increased during the period of the bubble economy in the latter part of the 1980s and declined during the recession in the late 1990s. It appeared as if the cause of this was an increase in early retirement, but in fact it was caused mostly by a decline in self-employed workers, who had bolstered the high labor force participation rate among older males. Looking at types of employment of elderly men, since the late 1990s the rate of part-time employment and day laboring among male employees (aged 60 - 64) has increased. In contrast, the proportion of fulltime employees has shown a downward trend since the recession.

When we assessed a desirable form of employment system for encouraging the employment of senior citizens, as a result of analysis based on survey sheets from businesses of the Survey of Working for Old-Age People, we found that corporate retirement systems compel employees aged 60 or above to leave companies. At the same time, however, we found that they guarantee employment up to the age of 60, and that the employment of senior citizens is subject not only to customs and other cultural factors, but also to the employment system.

In response, OHASHI Isao (Professor, Graduate School of Economics, Hitotsubashi University) made the following comments.

The presentation deals in a detailed and precise manner with the actual situation in Japan with respect to senior citizens' employment in both its supply and demand aspects. The methods used are solid as well. The content of the analysis can be expected to have valuable policy implications, and confirmation of the impact of raising the age at which employees' pension benefit is received is an especially pressing issue.

In respect of the labor-supply analysis, because the Japanese pension system is categorized by type of occupation, it may be necessary to conduct separate analyses for employees of private enterprises and for public servants.

In the labor supply simulation, if the benefit reduction under the Old-Age Pension for Active Workers is zero, why is the estimated employment probability for fulltime employment higher than in cases in which the pensionable age is raised to 65? If the rate of wages is the same, the difference between the two is whether or not one is eligible to receive a pension, and if the income effect is positive, then the employment probability of the latter should be higher, owing to the lack of pension.

For people in their late sixties, the system of benefit reduction under the Old-Age Pension for Active Workers was introduced, but this system is applied very loosely, and I wonder how many high-income elderly people there are to whom this system can be applied.

What is the significance of the finding that shows higher employment probability when spouses are employed?

In the analysis of labor demand, that fact that negative effects are shown, as expected, on senior citizens' retirement from enterprises taking measures that target older employees appears to be a problem innate to such enterprises. This could happen through two routes. (1) Enterprises that try to employ senior citizens in large numbers tend to take such measures. (2) Economies of scale apply to education and training, and so the more senior citizens there are, the stronger the effects. Also, analysis on an industry-specific basis should perhaps be considered.

Numerous companies employ senior citizens, but the proportion of companies that take measures targeting the employment of the elderly is very small. Why are so few companies taking measures, in spite of the increase in the number of senior citizens and the effectiveness of government policies? What are the limitations and problems affecting each policy?

Prof. Higuchi and Prof. Kurosawa responded to these comments as follows.

Considering the occupational-pension character of the Japanese pension system, in our analysis we targeted only people employed by private enterprises at the age of 55.

With respect to the comments on the results of the simulation, it is essential to take account of the difference in employment choice between fulltime and part-time. That is to say, by making the reduction amount under the Old-Age Pension for Active Workers zero, the proportion of part-time employees is reduced and that of fulltime employees increases, so it has a bigger positive effect on fulltime employment. On the other hand, as the raising of the pensionable age has an impact on whether employment is continued or not, it should increase the numbers of both fulltime and part-time employees.

We will confirm how many people in their late sixties face a reduction under the Old-Age Pension for Active Workers.

As for the higher employment probability observed in cases in which spouses are employed, this could be interpreted, after allowing for the household income effect, as indicating that both people are involved in the decision-making as regards taking employment.

The reason why we did not conduct an industry-specific analysis of labor demand is that we sought to avoid difficulties in the interpretation of the results by taking the wage curves used as explanatory variables on an industry-by-industry basis.

Company measures for hiring senior citizens were adopted actively in the 1980s when there was a shortage of labor, but in the 1990s, when there was a labor surplus, companies lost their enthusiasm for stimulating the employment of older people. Companies must devise new policies in line with changes to the employment system and the pension system.

Q: The elimination or postponement of retirement will surely have an impact on the employment of youth. How can equality of employment opportunities be coordinated fairly between generations?

A (Prof. Higuchi and Prof. Kurosawa): In Europe in the late 1980s, youth unemployment became a serious problem, and policy measures were implemented with the aim of encouraging early retirement and securing employment opportunities for young people. In that way European countries succeeded in reducing the labor supply of senior citizens, but youth unemployment persists to this day. Since the categories of work undertaken by older people and the young differ, it is not clear whether substitutability exists. For detailed analysis of this it is necessary to use microdata.

The second presentation was by RIETI Faculty Fellow Shimizutani Satoshi titled "The Reality of the Elderly Viewed from Preliminary Data."

The Japanese population is aging rapidly at present, and by 2050 the population over 65 years old is expected to reach 36%. In order to take social security measures to address this situation it is necessary to analyze the social situation. In the past, analyses of pension, healthcare and elderly care systems were conducted separately, and were primarily analyses of costs and benefits from a macro perspective, but I believe that in future it will be necessary to conduct analyses based principally on microdata. It is hoped to develop Japanese versions of the HRS (Health and Retirement Study), ELSA (English Longitudinal Study of Ageing), and SHARE (Survey on Health, Aging and Retirement in Europe).

The available data in Japan have many limitations. For example, the National Survey of Family Income & Expenditure is rich in data in terms of items relating to household economic circumstances and expenditure, but there is no health-related information. The Comprehensive Survey of the Living Conditions of the People on Health and Welfare has ample data concerning demographics, health, and economic conditions, but the survey is conducted at intervals of several years, and the data is taken from different individuals. The survey by the Tokyo Metropolitan Institute of Gerontology is panel-data-based, but contains no items with questions on elderly care and consumption. Similarly, the survey by Nihon University is panel-data-based and includes numerous variables concerning health, but little information on income and savings. The compilation of panel data scheduled to start in November 2005 by the Ministry of Health, Labour and Welfare is a large-scale survey covering 40,000 people in their fifties, but it lacks detailed questions for understanding older people's behavior.

The purpose of the above surveys is to cast detailed light on the elderly's behavior, but they are deficient in two respects: they have insufficient information for evaluating social security policy, and the original datasets are not made public. The necessary information for evaluating social security policy includes information on (1) the role of public pensions in the elderly's lives, (2) the socioeconomic factors that determine health outcomes, (3) the boundary between public and private nursing care, (4) income and consumption trends, and (5) factors stimulating elderly labor supply. In order to implement social security policy suitable for the elderly it is essential to obtain data that includes this kind of information, and this is why Japanese versions of the HRS, ELSA, and SHARE are required.

The survey being conducted by RIETI, which could be described as a Japanese version of the HRS, ELSA, and SHARE, is currently taking the form of a pilot survey. In advance of the survey, Ichimura Hidehiko, Shimizutani Satoshi and Noguchi Haruko visited persons in charge of the HRS, SHARE and ELSA and were provided with technical support. As a result, the survey is to be based on SHARE and adjusted to suit its implementation in Japan. The pilot survey was conducted in December 2005, targeting 700 households in Ota ward in Tokyo, and the collection rate is currently around 30%. The questions are diverse, covering individual preferences, employment, health, memory and cognitive ability, expenditure on medical care and elderly care, income and pensions, arm strength tests, family transfers, information on housing, consumption, savings, assets, and social participation. As a result, various interesting facts have been noted, and yet these facts only show correlativity and do not cast light on cause-and-effect relationships. In order to clarify those cause-and-effect relations, larger datasets will be needed in the future.

Responding to the above presentation, Takeishi Emiko (Head Senior Analyst, Social Development Research Group, NLI Research Institute) made the following comments.

The current situation in Japan shows the lack of accumulation of panel data compared with Europe and the United States. The available panel data are limited to such surveys as the Longitudinal Survey of Middle and Old-Age People by the Ministry of Health, Labour and Welfare and the Living Standards and Life Planning Survey and the Panel Study of Middle and Old-Age People conducted by the NLI Research Institute, but their content is limited. In this Living Standards and Life Planning Survey the same individuals were surveyed every two years for the eight years from 1997 and 2005, and currently the data up to 2003 are available. Among the initial 1,500 samples in the first survey in 1997, the number of surviving samples up to 2003 is 814. The survey posed questions concerning employment, health, assets, and perceptions in daily living.

Questions about older Japanese people include their high employment rate, their high suicide rate, and their allocation of living costs after retirement, and I would hope that the RIETI panel data will provide some answers to these.

Panel data are very useful, as they allow dynamic analysis of the impact of policies, and that will enable us to analyze from a micro perspective the changes in the social security system and pensions. Another benefit of using panel data is the elimination of the heterogeneity of individual samples, which is extremely important in terms of controlling for the diversity of elderly people.

Important for grasping the actual situation of older people is retrospective data concerning their life histories and the quantity and quality of their social networks.

Crucial for producing data is the representativeness of the data, the maintenance of samples, and the careful handling of private information. In the case of senior citizens it is necessary to devote particular effort to sample maintenance, owing to possible problems such as illness or institutionalization. I hope that the data are made public so that large numbers of researchers can access them.

Next, Olivia S. MITCHELL (Professor of Insurance & Risk Management, The Wharton School of the University of Pennsylvania) made the following presentation titled "Informing Aging Policy: Microeconomic Panel Data on the Older Population."

The U.S. Health and Retirement Study (HRS) is a panel survey targeting people in their fifties and above, and has been conducted every two years since 1992. The sample number is 22,000, and the style is an interview for 90 minutes each with a husband and wife. As it contains many questions essential for policy analysis, for example concerning economic and labor market status, health, and family transfers, it has become an information resource with a huge impact on social security policy, healthcare, Congress, and researchers. The HRS provides useful information for the analysis of trends among senior citizens. Data currently available in Japan have many drawbacks, such as incomplete ranges of questions, insufficient numbers of samples, and the fact that they are not in panel format. Detailed data similar to that in the HRS are necessary for Japan, given the rapid aging that is under way. If a survey similar to the HRS is carried out in Japan, attention must be paid to the survey method, the number of samples, its reliability, and the timing of the disclosure of the survey. Of particular concern is that a panel survey in particular, by its very nature, requires a great deal of time for building up data. Surveys of this kind are currently under way in countries such as Chile, Australia, Korea and the United Kingdom, and it is a great pleasure to see that Japan will also be implementing one.

In response to these comments, Prof. Ichimura replied as follows.

To Ms. Takeishi

You mentioned the necessity for data regarding the jobs that are engaged in for the longest periods, but owing to limitations on the timing of questions we currently gather information on jobs at the time people are 50 years old. The same applies to spouses. We gather samples based on residents' registration cards, so people who move from their residences into institutions are omitted. In the case of older people there is a serious problem of sample bias, but the design of our survey is such that by acquiring panel data, this problem is being reduced steadily. For the overall planning, the pilot survey is currently being carried out in three places, and I believe that it will be acceptable if we can ultimately get a sample size of 10,000. In order to prepare a framework for social security policy, we would like to designate 20 locations and to gather 500 samples at each.

To Prof. Mitchell

The collection rate is currently 30%, and I want to increase that. I gather that for SHARE each country's collection rate is between 40% and 60%, so I would like to set our target at 50%. I imagine there must have been a very great deal of effort behind the HRS collection rate of 80%. A surprising thing about the HRS survey is that you have been able to collaborate with the government agencies related to social security; their endorsement must be a great help. That is very helpful for enhancing the accuracy and trustworthiness of data, and I wonder whether it will be possible in Japan, too. Owing to time limitations on this occasion we have conducted the survey on paper, but in future I would like to use computer-assisted methods.

Q (For Prof. Ichimura): When you pose questions to senior citizens up to 85 years of age, how do you confirm whether or not their answers are correct?

A: We conduct an interview-style survey, so unless a researcher makes an error in writing, I believe that there should not be any mistakes.

Q (For Prof. Ichimura): In creating panel data, do you liaise with any government agencies?

A: In the early stages we contacted the Ministry of Health, Labour and Welfare and inquired about the possibility of conducting the project together. However, as they were already conducting another survey involving the collection of panel data in a different form, and there were limitations with respect to the system, I gained the impression that it would be difficult.